5th International Conference on Machine Learning for Networking (MLN'2022)
Keynote: Iman Hmedoush, Nokia Bell Labs, France
Title: Reinforcement Learning for Irregular Slotted Aloha (IRSA) with short frames
Abstract
One of the main difficulties to enable future scaling of IoT networks is the issue of massive connectivity. Recently, Modern Random Access protocols have emerged as a promising solution to provide massive connections for IoT. One main protocol of this family is, Irregular Repetition Slotted Aloha (IRSA), which can asymptotically reach the optimal throughput of 1 packet/slot. Despite this optimistic bound on the throughput that IRSA offers, the problem is not yet solved as the challenges in the non-asymptotic case, such as small frame sizes and decoding errors, need to be handled.
In this talk, we explain the application of some Deep Reinforcement Learning algorithms to IRSA protocol that aims to enhance its performance. We present two variants of IRSA that have integrated Deep Reinforcement Learning algorithms. The first is Deep-RC-IRSA, a Deep learning-based IRSA approach with random codeword selection, where each codeword represents the transmission strategy of a user on the slots (i.e., defining precisely on which slots the user will transmit, instead of just deciding how many replicas will be transmitted). We also consider an IRSA scheme in the non-asymptotic case, with short frame sizes. Then, we introduce sensing capability to the nodes, so they can sense the environment and send jamming signals to learn how to interact and avoid collisions. Our initial goal is to to learn to interact, in the sense of learning a sensing protocol entirely through Deep Learning. Our proposed protocol has achieved an optimal performance of almost 1 [decoded user/slot] for small frame sizes and with enough sensing time (minislots).
Biography
Iman Hmedoush received her bachelor’s degree in Telecommunication and Electronics Engineering from Tishreen University in 2014. She received a Master’s degree in Telecommunication Engineering from Sup ́elec, France, in 2018. She received her Ph.D. in Telecommunication Engineering in 2022 from Inria and Sorbonne University. Currently, she is a Radio Research Specialist at Nokia Bell Labs, Paris. Her research interests include Modern Random Access technologies for IoT and Cellular networks, protocol design for the MAC layer, resource management, and Machine Learning application in these fields.